JeffJournal

What, you expected some kind of regular entries? Pshaw.

As usual, my travel has interrupted my regular flow of journal entries. Here's on that summarizes the last week or so...

Flew back to Philly to my 'rents place with Tracy for Thanksgiving. We flew via Cleveland which was apparently getting snowed in when we arrived. So we diverted to Cincinatti and came back to Cleveland before being able to land. Luckily, the good folks at Continental were able to get us on another flight to Philly that night, so all was well.

Had a big clan gathering at my sister's in Allentown the next day, which was pretty cool. Everyone was there with the exception of my cousin Maggie. The PlayStation 2 was killer, too. I whomped my younger cousins at Tekken Tag Team, too, which was very cool, 'cause they have traditionally been much better than me at video games (go figure). "Who's your daddy?!?!", "Hey Chris, let me show you how to DIE", and "You're so weak, your momma tried to give you up for adoption and the Lemming family wouldn't even take you" were all popular phrases during this session.

We watched the traditional annual showing of "Airplane"; a Squyres classic. Brilliant movie.

• "Give me Ham on five, and hold the Mayo."
• "No thanks, we gave at the office."
• "The cockpit? What is it?"
• "That's when my drinking problem started."
• "It's a damn good thing he doesn't know how much I hate his guts."
• "Looks like I picked the wrong week to quit sniffing glue."
• "I've got to concentrate... concentrate... concentrate..."

Spent part of Friday working on mom's computer at home hooking up DSL to it. Part of the problem is that mom's Windoze installation is really broken somehow. It takes over 6 minutes to boot (i.e., to get to the Windoze login popup). It seems like it's timing out while looking for something during bootup, but I never figured out what it was. The first time I installed the DSL software, it was really flakey. It's with Verizon DSL, and they use this weird (IMHO) PPP-over-ethernet stuff. So you still have to "dial up" to get connected to DSL. And the IP address comes over that, too, so it's not regular DHCP. This kinda killed my plan to hook up my Linux laptop to their DSL and do things like check mail, etc.

There might well be some PPPOE Linux software out there; I haven't had a chance to check yet.

Anyhoo, I got it working more-or-less properly, but it didn't help much that the hard drive on that machine is failing. Every time I ran scan disk, it would find more bad clusters. Not good. The PPPOE installation finally go so flakey that I removed mom's pre-existing Netscape and the whole PPPOE installation and started from scratch (gotta love the non-deterministicness of Windoze!), and that seemed to make it much happier. I installed Zone Alarms firewall, too, which was kinda neat. Since they don't have a fixed IP, and aren't "connected all the time", the chance of an attack are smaller, but are still there, so I installed it. It's not perfect, but it's not a bad firewall.

I was going to spend some time on Saturday trying to figure out why it takes so long to boot that machine, but I caught some weird 18 hour flu that's going around the north east right now, and it killed me for the whole day. Tracy and I were supposed to go out to a nice dinner that night, so that didn't happen, either. Bonk. But the Irish had a convincing win over USC to round out our season. So it's probably a pretty good chance that we'll go to a bowl. Woo hoo! I've been hearing Fiesta, but I haven't been following it too closely.

Dad and I used the small business pricing from Dell to order a new Windoze computer for Tracy (she currently has a P1-133 with 24MB of RAM, which is painfully slow) under the "Wayne True Value" auspices (since my dad owns a small business, they apparently don't check, but it is legal since my dad bought it, and I do lots of consulting for him). It's actually supposed to come today (1 Dec), according to UPS.

I also got plane tix back to Philly in about 2 weeks to install DSL at his store. He's got a LAN of machines that need to be adjusted and whatnot such that DSL will be safe to install (need to change all the IPs, harden up the unix server a bit, etc.). I won't be able to use linux as a firewall (my initial plan) because of PPPOE issue, so I'll have to use Windoze 98's internet connection software (blech; although I will be looking for some Linux PPPOE software so that I don't have to do this).

Tracy and I flew back on Sunday morning without any major incidents.

I drove up to ND on Monday morning to get there in time for Kevin Barker's MS defense. It was all about the percolation model. Pretty neat stuff. It still performs poorly right now, but it's still in the early stages of development. He passed! Whoo hoo!!

Went to dinner with Kevin, his parents, and Shannon. It was good to see Shannon again; she's funny. She's also starting a PhD program (this Spring, IIRC) in Ohio; rock on! Dinner was good (at Basil's). Then I went back and hung out with Suzanne and Ed and watched a few episodes of Level 9. Not a bad show; it's got a good mix of techno geek stuff and action.

Kevin turned in his thesis the next day, and ended up staying that night as well, so a bunch of us took him out to the Mishawaka Brew Co for a few beers: Mike N, Dog, Jeremy, Ron, Shannon. Great conversation all around, and lots of laughs. Good to see/hang out with Kevin again. Perhaps he'll come back to ND; that would rock.

@#%#@%#@$ I had forgotten to fill out my reimbursement form for SC2000, so I had to wait until the CSE offices opened in the morning before I could leave to drive back home. I had to drive straight to my church where I'm doing some volunteer consulting with them for their various computer things (as I think I've mentioned before, they have a LAN with about a dozen windoze machines on it). We talked some more about DSL (we're putting it up to the budgeting committee in about 2 weeks), ordered a site license for Norton Anti-Virus, and discussed a few other random things. The anti-virus media should arrive in a few days; we planned on me coming back next week to install it on all the machines.

Planning for DSL takes a surprising amount of details:

• Moving their web site; it's on some local Louisville hosting service now, but DSL comes with 20 free MB of web hosting space
• Moving their DNS name for the same reason
• Re-training their web masters to use the new location, not the old location (should be easy, but...)

• Change the IP addresses that each machine has; I think they're random right now. I'll have to change them to be 192.168.x.y or 10.x.y.z or whatever.
• They use AOL for all their mail now; this DSL service comes with 20 free mailboxes.
• Changing all their e-mail addresses to be @churchofepiphany.com (and decide what the format of the e-mail IDs will be; an internal policy decision for them).
• Change everyone over from AOL browser/e-mail to Netscape/IE (haven't decided yet) and some mail client, likely Netscape/Outlook/Outlook Express (haven't decided yet). This will involve re-training everyone.
• Ensuring that everyone's address book and web bookmarks can be snarfed from the AOL software to the new software.
• Setup/ensure that the dialup works for the one workstation that they have off site (this DSL provides a free dialup for remote users).

• Shut down the AOL accounts fairly quickly after this all happens to prevent the two-email-address syndrome.
• Shut down the Juno account that the off-site user is currently using for the same reason.

As a good engineer, I have to document everything that I do for the above. Most importantly, however, what needs to be documented is the firewall configuration. This DSL service comes with a Netopia router which can also act as a DHCP server and firewall. It's supposed to be easy to configure, but we'll see. This needs to be documented because I won't be there forever.

Some other projects that they may wish to investigate after the DSL stuff gets all happily installed (probably mid-late January):

• Shared fax on the LAN (should be easy, I think).
• Group scheduling of resources (conference rooms, the community center, etc.).
• Random training classes, perhaps even some "intro" and "advanced" kinds of classes.

Had all my four of wisdom teeth out yesterday, as well as one more molar that wanted (he wanted to keep hanging out with the wisdom teeth, apparently). I was knocked out for the procedure. I think I pseudo-surfaced in the middle of it, 'cause I felt some rather strong forces (not pain, just pulling, etc.) on the right side of my jaw. It only took about an hour, actually. Apparently, my upper right wisdom tooth gave them a few problems (nothing major), but everything else went fine.

My jaw was fairly sore all yesterday; they gave me some mild pain killers and some antibiotics so that nothing gets infected. I go back next week to have my stitches removed. All in all, it wasn't nearly as eventful as I thought it would be (I guess I expected much more pain). My jaw is still fairly sore, and I'm not back on solids yet (checking is somewhat painful), but that's supposed to go away in a few days.

It's funky, though -- I can feel the end of the line of teeth with my tounge where that last non-wisdom molar used to be (on the upper left). So I can feel the end of my tooth line, which I have never been able to do before. Funky.

I'll be heading back to ND next week to visit with Bemen Dawes (sp?) from the Boost group. He's coming to visit with Jeremy, Rich, and Andy. I'll tag along for usability and other kinds of user-concerns, but probably not too much in the design and other stuff.

That's about it for now. Gotta get back to work...

I have to admit, I've never seen it actively snowing in Louisville.

It's not really accumulating, but it is kinda nice (I'm a cold weather person).

My jaw is doing ok; a bit sore, but manageable. I actually managed to have a few slices of pizza last night.

Tracy's new Dell 800Mhz came yesterday, and I spent a good amount of time setting it up. Copied a lot of stuff from her old Windoze machine to this one (it was a pain in the butt to export/import her addressbook and message folders from Outlook 98 to Outlook Express 5.5.1 [Outlook 98 isn't available anymore, and Outlook 97, which I have on CD, doesn't do IMAP]). Finally got everything over, though.

One annoyance, though -- Outlook Express has some nice rule-filtering capabilities such as "take messages with such-and-such subject and automatically put them in folder foo". Very handy. But it doesn't work with IMAP inboxes! Why not?!

Got the latest distributed.net client and installed it on there, too. But it only seems to want to do RC64 --
it simply won't do OGR. Weird!

The new version of mojonation sucks. It keeps coming up with an error than causes it to lockup. That is, it only runs for about 5-10 minutes and then locks up (something to do with bad XML parsing). Woof. I also wonder why they don't use their mojonation-announce list to announce new versions; it's advertised on their site, etc., but I never get announcements from it. I only get new versions when I happen to notice them. Weird.

Just a few quickies today. Gonna spend some time on the ogg vorbis encoder and LAM today (-pty fix in Linux and still trying to get some reasonable fault tolerance issues worked out).

These are interesting times that we live in.

A series of random things have been occurring. Hence, this will be a random journal entry, written while I eat my lunch.

I have discovered that the HTML element <HR> is not always centered by default, particularly when it is less than 100%
of the width of the browser. I don't know if this is specified in the HTML spec or not, but I have found that KDE's Konquerer browser (which actually isn't a bad browser, surprisingly enough!) does not automatically center the following:

which means that all thousands of screaming jeffjournal fans out there that are viewing my journal archives in Konquerer are wondering why there are half-line separators on the left in their browsers. Oops.

I have now mended my ways, and write dramatic half-line separators like this:

All is right within the world.

Along the same lines, it's quite tiresome to type out "<CENTER><HR WIDTH=50%></CENTER>" (particularly when you have to escape it to write it in example form so that you can read the "raw" HTML in HTML). So I think I need to add some special "escapes" to the jcc such that things like this are automatically done for me.

I'm thinking of escapes for:

• Dramatic half-line separators

• URLs will automatically be linked

• In a wikki/doctext-kind-of-way, make lists easier (where "easier" == "use some abbreviated syntax that jjc will expand into the correct HTML for <UL> / <LI> / </UL>, etc.")

• In a wikki/doctext-kind-of-way, make linking be easier (similar definition of "easier" as above)

• In a wikki-kind-of-way, make <code> and <strong> and <em> be easier, 'cause I use them all the time.
  

Also, Lummy has managed to get SourceForge running on our web server. He claims that some of the features in their diary stuff are superior to jjc (not surprising). However, I might have to steal some of them and put them in jjc, since I've kinda grown attached to it.

Stealing such features, however, has a fairly low priority.

I'm actively working on a parallel ogg encoder (stole the oggenc code, adding a whole new parallel personality to it, and renamed it to poggenc -- oggenc effectively has a brother now). I re-read my white paper on generalized manager/worker using both threads and MPI (and wow, it was long!) to remember all the thoughts that I had about that. Found a few minor errors, and was annoyed to discover that my formulae at the end were just about entirely wrong. Math sucks.

So I've coded up a bunch of the framework so far, and have started classes for the input, worker, and output threads. I've added the necessary #define's for thread safety within Vorbis (which is isn't yet, in the way that I need for this -- it can't handle multiple threads simultaneously encoding on the same stream), and #define's for MPI. Seems to be going well.

Happily hacking
Abstract turns into concrete
Parallel vorbis

Hacking hacking hacking...

I was manually archiving the web logs on www.lsc.nd.edu yesterday (really gotta finish automating that process someday...), and I did the normal "bzip2 combined_log". I did this on a Hydra node (400Mhz UltraSPARC II). After a good many minutes, it didn't show any sign of finishing (the logfile was approximately 189MB).

I had anticipated it to take a while, but it was actually taking longer than I expected. The idea popped into my head: "I wonder how much faster the new Sun UltraSPARC III would be able to do this!" We actually have a SunBlade (750MHZ ULTRASPARC III) on loan from Sun (shhh!!!), so I copied the log file to its local disk and started a bzip2 of it (Solaris 8 seems to ship with bzip2 -- rock on).

After 50 minutes, the SunBlade finished. The Hydra node looked like it was about 1/3 of the way finished. Ouch. I then though, "uh oh -- I don't know if these two versions of bzip2 are the same; am I comparing apples and apples?" So I checked. Oops -- the Hydra was using bzip2 0.9.0b and Solaris 8 (the SunBlade) had bzip2 0.9.0c -- both of which dated back to 1998.

So I went out and found that the current version of bzip2 is 1.0.1. I downloaded it to both machines and compiled it with "-fast -xarch=native -xtarget=native", and re-ran the test (both from the local hard drive, of course).

The SunBlade finished in 7 minutes flat. The Hydra node finished in about 15:30. Wow.

Morals of the story:

• The bzip2 that we have (had) on AFS sucked. I recompiled the new one with optimization and put it out on AFS. I got the OIT to update theirs (Solaris 7 tree), too.

• The bzip2 that ships with Solaris 8 sucks.
  
• The SunBlade was slightly more than twice as fast as our UltraSPARC II. But that only naturally follows, 'cause its clock speed was almost twice that of the Hydra node.

Just a few interesting data points, nothing more.

In Lummy's never-ending quest for good web collaborware, I found a bug-tracking system called RT, that seems to be a web-ified version of ANL's req system.

It seems to be pretty nice -- it has a bunch of features without being overly complicated (ever had a look at bugzilla? I can't even understand how to use that thing!). It doesn't do everything, but it seems to do most of what we need. Most importantly, IMHO, it has an e-mail interface (something that Jitterbug lacks), so the admins don't have to go to the web page to do quick-n-dirty bug tracking things.

I set it up on my router and let some of the guys in the lab play with it. General consensus was that it wasn't bad. I tried to get the CVS copy of RT going (has a bunch more features than the current stable release), but it seems to be not-quite-ready for prime time yet. We'll have to wait for that, I guess. :-)

Lummy said he might try and tie RT into the SourceForge that is running on lsc.nd.edu, but it's more likely that he'll just make some kind of primitive e-mail interface to the bug tracking system that is already in SF. Or, it's more likely that we'll all just bitch about it and nothing will get done. :-)

I noticed something annoying about the CVS version of Vorbis. Some background...

Vorbis is the music format. Ogg is the file format. That is, you pack vorbis data into .ogg files. There are separate libraries to do each. Additionally, there's a third library do write output to sound devices called ao. So to compile oggenc (the ogg/vorbis encoder), you need to configure it with:

This seemed pretty silly to me, especially since you typically install all three libraries and oggenc into the same place. So I hacked up their .m4 files to check the $prefix if the corresponding --with-* option was not specified, and submitted it to the voribs-dev mailing list. We'll see if the patches are accepted (they were really only a few lines of shell script; not rocket science).

However, the vorbis-dev mailing list seems to currently be down. I see from the web archives that a few posts (including mine) have been sent since last Friday, but I haven't received any of them. I sent a few queries but have heard nothing back yet. Hmm.

Speaking of oggenc, I have now done much coding of a parallel version (the part above about working on poggenc was written yesterday; I just haven't submitted this journal entry yet :-). I have been following the design laid out in my white paper about mixing threads and MPI for multi-level parallelism, and it seems to be going well. Most of the infrastructure is done, and I'm just starting to code up the parallel aspects (shipping the audio data to remote nodes, shipping the ogg data back, etc.).

When I have some semblance of a working copy, I'll probably ping Dan at Scyld again.

I obviously didn't go to ND as planned this week to meet with Beman. I hear his trip was a great success and many great things were discussed, but I simply couldn't do 9+ hours of driving this week over a 2 day span; it was just too much. I guess I'll meet him some other time.

Tracy's new 'doze machine seems to really chug through RC5 packets. And it has recently decided to start doing OGR packets (I couldn't get it to do OGR before; go figure). The distributed.net client seems to suspend itself unpredictably, however. For example, I left the computer on since yesterday for the sole purpose of RC5 hacking, and turned off the monitor. This morning, I turned it on and the last activity in the distributed.net log was from yesterday.

Weird.

I watched the SciFi channel's Dune saga, parts 2 and 3 (missed part 1). Not bad. I saw the original Dune movie a while ago, and I guess I rate these two as about the same. Some of the special effects were good (in the new one), some kinda sucked. I read the original Dune book, but none of the sequels.

But I'll probably watch the sequel movies when they come out; I enjoyed this version of Dune.

I went to have my stitches out today from the oral surgeon (had my wisdom teeth removed last week). No big deal there. The doctor, coincidentally, is an ND grad (he mentioned it when he saw my ND varsity jacket). I knew that I liked him for a reason.

More interesting, however, was my drive to and from the doctor's office. It solidified my understanding of "every action has an equal and opposite reaction."

The office is a few miles away, and I basically take one road to get there (a fairly main traffic artery). There are many lights between my apartment and the doctor's office. On the way to the office, I only had to stop for one red light. On the way back, almost all the lights were red.

Everything seems to work out evenly.

However, I noticed that the muffler is going out on my car. @#%@#$%@#$!!!!

And of course, in the 45 minutes that I was gone (which was the only time I left the apartment during the business day all week), I missed an Airborne Express shipment from outpost.com (free overnight shipping!) with Turbotax. Quicken/Turbotax is the only reason that I use a Windoze machine with any regularity. For those who don't, and if you happen to have a spare 'doze machine lying around, I highly recommend them. They're great products (I wish they had Unix equivalents; there's gnucash, which, as I understand it, is more or less like a less-mature Quicken, but no equivalent for Turbotax).

I have to admit, I'm not impressed by the latest Mozilla.

Within 3 minutes of installation, it managed to freeze my window manager. All I did was try to go to "Encrypt sensitive information" (one of the options off their drop down menus), and whammo. Or, I guess I should say, "...", because that's more like what really happened.

Additionally, I'm disappointed that it still doesn't come with crypto installed. The US export restrictions have all but been removed, so why not ship with it? SSL is necessary for a large portion of web usage these days (i.e., e-commerce sites), so why do I have to go through a second step to go get crypto and install it?

Additionally, the crypto that I installed only seems to work for root. Doesn't work for my regular user. <sigh>.

Saw a great article referenced from slashdot yesterday on the [lack of] software quality. I highly recommend reading it; it'll take 10 minutes of your time.

The vorbis-dev list is still down, which is annoying, 'cause I've got some questions about the DSP encoding stuff, and how to share boundary values in the persistent state...

Just saw a great quote from Donald Knuth (read "the man") via Ron Garcia (probably paraphrased, but so what):

Premature optimization is the root of all evil.

Some other random great quotes that are popping into my head right now:

• Welcome to the LSC. If this if your first night, you have to code.

Ok, that's the only one that's coming to mind right now. But it's a great quote, nonetheless.

Interestingly enough, my cell phone is acting a bit whacky. The other day, I found a few new features in it that I activated, including one for "audible alerts when switching between analog/digital." Yesterday afternoon, it was sitting in its charging stand over on the other side of my computer room, quietly minding its own business, when all the sudden, amid the clatter of the clicking keys of copious coding, "BEEPBEEP!" shrilled from that corner of the room.

Startled (and almost knocking over my soda), I whirled around to see what it was. Nothing was out of place. Nothing had moved. No telltale signs of destruction from a burglar who had somehow managed to sneak in, rip up my computer room, and leave (all in the span of a single "BEEPBEEP!") while my attention was focused on coding.

Weird. Chalk it up to sun spots.

Seconds later, "BEEPBEEP!" shrilled again.

This time, I did knock over my soda.

Swearing loudly, and cursing the Gods of the Phantom Beep, I endeavored to figure out what the hell was beeping. I acted like a hunter; standing quite still (in the corner of the room with my back against the wall), controlling my breath, only moving my eyes in an ever-watchful dance around the room. These tactics will trick the Beepbeep to come out in the open -- if I'm not moving, surely it can't see me.

Indeed, 3 hours, 14 minutes, and 27.2 seconds later, my skillful hunting prowess was rewarded: "BEEPBEEP!". My eyes quickly located the source of the sound, and were rewarded with seeing a quick change on the screen of my cell phone -- too quick to read. It must have heard my heart jump at its shrilling, and frightened the text on the screen back into the depths of its circuitry.

Immediately, I pounced -- intimidation is the only method that works here. "Say it!!!" I screamed. "Sayitsayitsayit!!!" (Sam would be proud). And the frightened cell phone meekly said, "beepbeep", and it paused the words on its face long enough for me to read them:

After this little "encounter", it took some time to heal the emotional wounds between my cell phone and I. But we're better now; I think the counselor is helping a lot.

Ok, here's another great quote:

"There are three types of people in this world: those who can count, and those who can't."

Yow. Been traveling. Journal suffers. Millions weep and gnash their teeth.

Until now.

Be at peace, gentle reader.

Went and saw the play "Rent" with Janna and Tracy last week. It's a good play, depressing and cathartic. I've seen it before (in London), but none of the others had. We were supposed to go to Tracy's work Christmas party afterwards, but the play wasn't over until about 10:30, and by the time we would have gotten there, it would have been over.

Went and had a beer with Janna afterwards, which is always cool. As Anna likes to remind me, "they're my only friends in Louisville" (which isn't far from the truth!). I haven't had much opportunity to get out and meet random people here in Louisville, but this doesn't really concern me. I think that after I get my Ph.D., I might try my hand at an adjunct faculty position at the University of Louisville or something -- something to get me in touch with the geek crowd down here. Who knows.

But until then, like I said, I'm not really too worried. I do have a small group of friends down here, and I've met several of Tracy's work colleagues; they're all nice folk.

Flew to Philadelphia at an absurdly early hour on Saturday morning. Drove from the airport straight to dad's hardware store and started to work on hooking his store's LAN up to Verizon DSL. This actually entailed several things:

• There is one SCO unix server on the LAN, 3 windoze 98 machines, and 6 DOS machines (!). This was actually mentioned in a recent /. article -- the DOS machines are fully functional; there's really no need to replace them. They use attractive ANSI graphics to do inventory queries, price lookups, etc., etc. 3 of those 5 DOS machines are actually cash registers, and are quite functional (and have been for many years). All 5 machines are very dependable. Sure, there have been a few random quirks, but for the most part, they have served remarkably well over the years, and continue to do so.

  However, whoever at True Value designed the network did so poorly. It seems like the IP addresses were chosen at random. As such, they were unsuitable for connection to the internet. So I had to convert all the IPs to be of the form 192.168.x.y (one of the approved private networking domains). Changing the Unix machine was easy. Changing the windoze 98 machines was also fairly easy. Changing the DOS machines proved to be a little more work -- their TCP driver is loaded dynamically from the DOS command line (it's a .SYS file loaded by a proprietary INET command for their TCP stack). But get this: the IP number, netmask, hostname, etc., etc., are all encoded in this .SYS file.

  I had to ressurect the memories of how to alter the .SYS files out of long-term storage (the percolation model, but several cycles were wasted during stalls while waiting for the memories to surface). But in the end, I triumphed. Actually, I have to hand it to those old DOS programmers -- once I remembered how to do it, it wasn't too bad of an interface for the time (it's all command-line driven).

  So I got everything on 192.168.x.y. Just for the heck of it, the Unix server was alone on 192.168.30.x, the 3 windoze 98 and 2 DOS "lookup" machines were on 192.168.20.x, and the 3 cash registers were on 192.168.10.x. Everything appeared to be working smoothly.

• First glitch: Dad opened up at noon on Sunday (special Christmas hours; he's not usually open on Sundays). At 11:59am, I get an intercom call from him, "The credit card functionality in the cash register isn't working -- it freezes up." Doh! I had tested cash transactions and they all worked fine, but I hadn't tested credit card transactions (the cashier swipes a card in a slot that it built in to the computer keyboard -- pretty slick, actually -- and the cash register makes some TCP or RPC calls to the Unix server [not sure which; I've never been privy to the internals of the True Value code] who has multiple modems that it uses to make the outgoing call to the credit card center, verify the data, etc. There's a little progress screen on the cash register [ANSI graphics, mind you!] during this time: "Looking for modem" / "Dialing" / "Sending" / "Waiting" / "Approved").

  It took over an hour to figure out what was going on. Actually, my dad spotted the problem without realizing it. While rebooting (one of several while we were trying to figure this out) to get avoid the "hang" produced by the faulty behavior, he said, "Hey look at this --
  one of these status messages that zips by quickly in the beginning flashed a negative number. Does that mean anything?"

  It turns out that the cash register was calculating its ID incorrectly. In the True Value system, cash registers are numbered sequentially from 1 (stupid Cobol programmers -- they must have married Fortran programmers!). Each physical cash register has a fixed ID that never changes. It seems that our 3 registers now thought that they had IDs of -17, -18, and -19. Doh!!

  So even though the main cash-register-processing-routines in the Unix server (every transaction is transmitted back to the main server in the back room) were happily accepting most transactions from negative-numbered cash registers, it seems that the credit card authorization routines were saying, "Hey -- you're a negative number cash register. This must be a mistake. Go away." And therefore the cash register would hang, because it would either not get a response from the server, or it would get an error response that it didn't know how to interpret.

  So that identifies the problem (always an important -- and frequently overlooked -- step). Now, what was the cause? The only thing that I had changed was the IP address.

  No way.

  Way.

  It seems that the True Value programmers calculate the cash register's ID number off the IP address. More specifically, if the IP address is w.c.y.z, the cash register's ID is (20 - z).

  No, I'm not kidding.

  The /etc/hosts file on the Unix server with all the original IP addresses had the cash registers starting with x.y.z.20. I had changed them to be 192.168.10.1, 192.168.10.2, and 192.168.10.3. Doh!! Fixing them up to be .20, .21, and .22 solved the problem.

  How fucked up is that?!?

• For the next part, for a long series of reasons that really aren't worth going into, we decided that my dad's windoze 98 desktop machine would be the DSL gateway into the LAN using Microsloth's Internet Connection Sharing functionality. So I first disconnected dad's machine from the internal LAN and brought up Verizon DSL on it. I installed a firewall, got everything working, etc., etc. They use PPP-over-ethernet as opposed to standard DHCP-style setup. Hence, it actually uses the Windoze dial-up networking functionality to establish a DSL connection to the internet. Really weird. I read some RFCs and position papers about this (PPPOE is actually standardized, and will be in the mainline Linux 2.4 kernel), but I still don't see the benefit. "You mean I still have to invoke kppp to activate my 'always on' connection?" It just seems weird to me.

  So I activated the Internet Connection Sharing (ICS) stuff, and noticed that it changed the IP address from 192.168.20.x to 192.168.0.1. Hum. I'll bet that was for a reason -- default router on a network, etc., etc. So I looked it up in the online help -- sure enough, it says that you can use addresses in the range 192.168.0.2 through 192.168.0.253.

  WHOA!! WTF?!? They changed a class C private network to a class D!! $%@#$%@#% I can see forcing the ICS server to be 192.168.0.1, but why the hell did they make a netmask of 255.255.255.0 instead of 255.255.0.0?!? That annoyed the hell outta me. And it had the following consequences:

  1. I had to go change the addresses on the other windoze 98 machines to be 192.168.0.x.

  2. Even worse, I had no DNS IP numbers to put in the Windoze 98 machines, since PPPOE "takes care of this for you" (yet another aggravating aspect of PPPOE -- DHCP can do this as well, but you can still manually override it on the client, if you want. You can't override it with PPPOE). Indeed, Verizon wouldn't give me their DNS server IP addresses for this very reason, "It's not supported". I managed to get them myself (whois and nslookup, no rocket science there), but they're not accessible from inside their internal DSL network. Arrggghh!!

  3. Hence, I had to fully surrender the windoze 98 machines to the ICS setup: I had to set the clients to use DHCP to get an IP address (apparently the ICS server turns into a DHCP server on the local network), disable DNS, and enable a setting labeled "Use DHCP for WINS resolution", which apparently did the DNS resolution stuff.

     I'm not sure what WINS is, but I thought it had to do with NETBIOS stuff. Apparently not...?

  4. What further sucked was that these machines now had an automatic netmask of 255.255.255.0. Which means that they couldn't reach the Unix server, because it was on 192.168.30.1 -- i.e., outside the netmask range, so they were sending packets to the default gateway who had no idea what to do with them and probably dumped them on Windoze's equivalent of /dev/null.
     So I had to put my unix server on 192.168.0.250. This is actually risky, because since the ICS server is a DHCP server, 192.168.0.250 is in the range of addresses that it is allowed to give out. Hence, I could have an IP address conflict. I can only hope that the DHCP server does the Right Things and sticks to low numbered IPs (there are only 2 clients, after all), and re-uses them when new DHCP requests come in.

     That just really burns me up -- that the stupid Microsoft programmers automatically assigned a class D netmask to a class C network. Only assigning 0.1 through 0.250 via DHCP would be fine, but the prohibitive netmask prevents [safe] interoperability with anything else that is not Microsloth on the same local network. I suppose that I shouldn't be surprised by this, but it still sucks. In hindsight, I should have done this with a Linux router and it would have been much easier and less time consuming. <sigh>

     MICROSOFT SUCKS!!!
     

  
  

I also had to reconstruct my mom's windoze 98 machine at home. It was basically so broken that it required a full reinstall. To make matters worse, the physical C drive had bad clusters on it, and every time you ran scandisk, it would find more bad clusters. Hence, it was going bad slowly. Not to worry -- we had a second physical disk already in the machine. So I just swapped the two disks (there are 2 IDE interfaces in the box; the boot disk was the master on one by itself, the second was a slave on a interface with the CDROM) and reinstalled on the old D drive.

Whoops -- the machine now takes over 5 minutes to boot (!). There were three distinct locations in the boot where it appeared to stall
-- waiting for something for which it apparently eventually would time out and continue. I only discovered today (i.e., after 2 days) that swapping the disks back to their original positions on the IDE interfaces (even though the boot disk is now different -- changed in the BIOS) eliminated two of the three delays in the boot time. I couldn't believe that this was true, so I swapped the disks and changed the BIOS settings back, and sure enough, it timed out in 3 places instead of 1. That's fucked up. The Intel architecture sucks.

And as for the last time out during the boot -- I have no idea what the heck that was. Dad has a total of 3 machines from this company, and the other two don't do this. It even did it after a fresh, clean install of Windoze 98, so it must be something in the hardware. That's fucked up.

Dad was a bit disappointed in the e-mail performance of Verizon's SMTP servers. It was really slow to send mail this weekend. Much slower than his old dialup account (56k). Indeed, several times it timed out and we had to click "send" again in his e-mail software.

A call to the Verizon help desk got a recorded message, "Verizon customers may be experiencing difficulty sending and receiving mail. We are aware of the problem and are working to fix it..."

It turns out that Verizon got heavily spammed. A /. article about it said that Verizon is convinced that it was deliberate and malicious. Apparently, they brought up more servers yesterday to try to cope with the load, but are still trying to automate the spam rejections.

Along the same lines, my dad does a lot of stuff for the ND Alumni Club of Philadelphia. One of the things that he does is send out mass e-mails to both the club members (it's a pretty active club, actually) and students on campus from Philadelphia (e.g., he passes along e-mails about rides home for students). Hence, he can send out an e-mail with several hundred BCC recipients.

His old ISP was a small local firm, and he got permission to do this. Verizon's max recipient count is 40. And especially in light of the spam problem they had this weekend, they were not interested in raising it at all. So I have to setup some special stuff on lists.squyres.com for Dad to relay his messages out. The problem is that he has a database with the e-mail addresses in it that he sends to, and it gets updated frequently (multiple times a week). So he needs an automated mechanism to import a whole new list of subscribers and completely ditch the old list. I'm thinking of some scripting with GNU mailman to make this work...

I hate Windoze. I am so glad that I don't have to use it on a daily basis. I can't imagine how people actually get work done with it. A very large company that I know (no names mentioned...) has their employees run "weekly updates" from the IT department on all their Windoze boxen (i.e., it runs automatically when you boot up). If it's Monday, it asks if it should run the weekly update. You can click on "No, please delay the update; do it later" up to 3 times. After that, it will run the update no matter what.

The updates routinely take around 2 hours -- if it runs smoothly, which they usually don't. The update frequently hangs/crashs in the middle of the process (your only indication of this is if you happen to notice that the hard drive light stops blinking for an extended period of time, upon which you have to reboot and start the update over again). You also can't use the computer during that time. They don't schedule them to run when no one is there (e.g., 3am), because to save money on electricity, everyone is required to turn off their workstation at night. Hence, this weekly update procedure is guaranteed to make their computer unusable for at least 5% of their ANSI standard work week (2.5% when the standard gets updated to reflect common practice).

I just can't imagine having to work in an environment like that.

On the drive between my dad's store and my parent's home, I saw a most curious thing: a cell phone tower that is disguised as a pine tree. It is painted brown and has evergreen branches on it. It's still a dead giveaway because it stands out much taller than any of the trees around it, but it did cause me to do a double-take.

I have to say: this election is working out exactly as rmurphy4 predicted. The latest bit: FL legislature picking electors.

Kudos to you, sir! And kudos again!

I swear that Perk's older brother works in the MailBoxes, Etc., here in Louisville.

It's either him, or someone that looks exactly what Perk will look like in about 5-7 years.

It is 4:53pm.

Dad's network is finally alive again.

And I don't know why.

At time T, we were at state A.

It didn't work.

We changed one thing, theoretically moving to state B.

This, of course, entailed a reboot.

State B didn't work. So we changed back to state A.

And rebooted.

Suddenly it worked.

And before you ask, I'm quite sure that we only changed one thing, and then changed it back. Yes, something changed during that time, but it sure as heck wasn't from something that we did -- Windoze did something internally.

I think that this is what bugs me most of all about Windoze -- its nondeterminism. It doesn't matter how smart you are, nor how much you know about computers: sometimes it works, sometimes it doesn't. There are absolutely no guarantees about consistent behavior in Windoze.

I'm just bitter at the end of a long, frustrating day where I got absolutely nothing done. <sigh>

The uptime on my desktop linux box is 98 days and counting.

I've been attacking the stack of mail in my inbox from my foray to Philadelphia.

Lummy asked me to step into the Boost discussions about directory structure and whatnot. They do not seem to understand the necessity of separating a source code tree from an installed tree. We'll see where this discussion goes.

Had scads of LAM mail, both from the list and people who mailed me individually. Plowing through all of that...

Got some good responses about parallel vorbis. Turns out that since vorbis is a differential encoding method, it will require the same technique that I used in parallel bladeenc -- sending some redundant input blocks to each processor in order to "build up state", so to speak. This is kind of a bummer; it means that the parallel output will likely not be diffable against the serial output. But some of the vorbis developers indicated that the general idea should be able to work. Hopefully, I'll be able to work on this later today.

Lummy bought me a webcam (Intel Personal Camera or something like that) for use in teleconferencing up with ND. I can use my headset with it, which is trez kewl. It's also detachable from the computer and can serve as a portable digital camera. It's not the world's greatest camera or anything, but it could prove to be useful. Brian and I had a few difficulties getting a netmeeting going on between squyres.com and nd.edu last night. Apparently, the fact that my Windoze box is on a private network behind my router is the problem. Brian/Pete found a kernel module for the NAT that should fix the problem, but I haven't been able to successfully compile/install it yet. We'll see how that goes; could be really useful in terms of communicating with the Home Office.

Windoze really really sucks. My dad, after I got DSL all working and whatnot -- his desktop computer is the Internet Connection master for the local LAN in his store, i.e., DSL comes into his machine and is routed to the rest of the local LAN from there -- was having problems with a tax program that he uses to pay taxes on the salaries for his employees. It's supposed to dial an external phone number through the modem and then Do Its Thing. But for some reason, it didn't work last night when Dad tried to do it.

So Dad called tech support for this program this morning. Their advice? "Yeah, we've had problems with people who have DSL -- even if you still use a modem, our software doesn't seem to connect properly. Let's try something; go to the control panel, network icon, and remove the entry 'Dialup Adapter'..."

Needless to say, this fucked up everything!.

A bit of background: Verizon DSL uses PPP over Ethernet (PPPOE) instead of normal DHCP stuff. PPPOE uses dialup connections to establish connectivity (it's weird; supposedly "it's easier on the user, because they already understand the dialup-to-connect concept". I think it's just stupid -- DHCP can do everything that PPPOE does, and not have to go through an additional dialup step). So when the tech weenie had Dad delete the dialup adapter, that totally fucked up the DSL connectivity.

I've now been on the phone with my dad for over 3 hours and it still doesn't work yet (we've finally managed to get the DSL connectivity back, but the internet connection sharing to the rest of the LAN is not working again yet). Windoze absolutely sucks.

I'd be willing to bet that the first few steps of the tech support checklists in just about every Windoze-based support center go something like this:

1. Ask the following question: "Have you rebooted your machine since the problem started?"

2. If the user answers no, have them reboot and move on to the next caller.

3. Ask the following question: "Have you uninstalled and reinstalled the product that you're having a problem with?"

4. If the user answers no, tell them to uninstall/reinstall and move on to the next caller.

5. ...

Windoze absolutely sucks.

It was a sad day yesterday.

Queeg, my linux desktop here at home, had its 100-day uptime destroyed by a hair dryer that tripped the power breaker.

There was much wailing and gnashing of teeth.

But I might use the opprotunity to upgrade queeg to Mandrake 7.2.

Do deee...

I finally got my muffler replaced yesterday.

That was $140 I didn't want to spend. Ugh!!

A series of random thoughts:

• Getting my muffler fixed (actually, it's the pipe between the engine and the muffler, hence it wasn't covered by the warranty) and running a bunch of Christmas errands took a good portion of my day yesterday.

• I talked to Darrell on the phone and found out that his parents live just a few miles from Tracy's parents down in Florida. It's a small fricken' world!

• I love the ability to track my UPS packages on the web.

• For those of you who keep bugging me, yes, I did screw up on a previous journal entry -- I meant to say class B/C networks in the entry about DSL and MS's internet connection sharing, not class C/D. It was late, and I was tired when I wrote that entry.

• Turns out that MS netmeeting doesn't work between two different private networks. i.e., from my 'doze box behind my router (which has a 192.168 address) to another box behind a different router (which also has a private IP address, perhaps of the 192.168 variety). It's not a shortcoming of netmeeting itself, per se, it's a shortcoming of the underlying protocol -- H323. Bummer. However, with a special H323 NAT module in the IP masquerading stuff in Linux, you can do a netmeeting between my private windoze box and somewhere else directly on the net -- as long as the box on the private net is the one who initiates the call -- which is quite handy.

• I wrote a schload of Christmas cards yesterday (will send them today) along with our obligatory Christmas letter. I'll post it here in a few days; gotta wait for people to get the snail mail version first. :-)

• I wonder what it would be like to own an elephant.

• Darrell has a great picture of his decorated house at http://christmas.kresge.com/. Isn't running your own DNS cool?

• Louisville has gotten 3 snowfalls of about 2 inches each. Each one has sent the city into a panic. It's quite funny to watch.

• Lummy and I are fixin' to go to Madison to meet with the Condor folks in early February. I sent some dates up to Erik, and he's checking into it.

• Progress on poggenc is coming along swimmingly.
  

That's about it. Back to coding. Mmmm.... coding....

Oops -- got the URL for Darrell's decorated house incorrect.

What the hell has happened to quality control these days?

Several cool things happened yesterday.

• I re-released PSR
• Brian re-released inilib (ok, that was actually 2-3 days ago)
• My sister got engaged (no URL available :-)

PSR is the Password Storage and Retrieval system that we use with OpenPBS to get AFS authentication with PBS jobs.

We've had problems with our installation of PBS over the last several months (ever since we upgraded to OpenPBS, actually). It turns out that one of the components of PBS, the Mom (a daemon that runs on each compute node and manages the user jobs that are launched on it) was at fault.

Actually, it was our patches to the Mom that were at fault. We had to patch the Mom to include bits to launch some PSR kinds of things (first, a program to get an AFS token, second, a program to "shepherd" the user's job and re-up the AFS token before it expires. Doing this allows a user's job to run for much longer than the life of their token -- their token is magically renewed for them for the entire life of their job).

We used the popen() call to invoke these two commands. Unfortunately, we didn't think that popen() would have the child process inherit the open file descriptors from the parent. But it does. Doh!!

Specifically, the Mom has multiple sockets open, including one that it is listen()ing on. To make a long story short, having multiple processes open sharing the same socket is a Bad Thing, and it caused Ickyness in PBS's runtime because it typically disrupted PBS's internal protocols.

Adding the following code to the beginning of the PSR executables solved the problem:

for (i = 3; i < sizeof(fd_set) * 8; ++i)
    close(i);

However, I still think that this is not perfect -- not knowing the internals of the Mom, I think it is still possible to get a race condition where Badness can occur. This can happen if the PSR executable is launched and them some Event happens on the socket before the PSR executable is able to close it. I think the real solution is to make the sockets be close-on-exec in the Mom, but I'm not sure. I've mailed the PBS guys to see what they think.

If you don't already use inilib, you need to. It will save your life! I classify it in the same category as the STL -- you could write something to do the same thing, but why?

inilib is a C++ library that reads and writes .ini files. While this in itself is unremarkable, its cool aspects include:

• Simple 2D array-like accessors. For example:
  foo["section"]["key_name"] = keyvalue;

• Small API; easy to remember and use
• Automatic write-upon-destruction semantics (if desired)
• Script-like automatic type conversion semantics. This is truly cool. By abusing some of the properties of C++ on the back end, we can do things like this:

    int i = 37;
    foo["section"]["some_integer_key"] = i;
    string s = "1000";
    foo["section"]["some_integer_key"] = s;
    i = foo["section"]["some_integer_key"];
    // i now equals 1000 

  More to the point, you can use the inilib objects like Perl or PHP objects -- all the type conversions are automatic and safe. This is utterly cool.
  

So anyway, go start using inilib. You'd be surprised how often you want to save a config file from your program; inilib just works.

My sister got engaged last night. It was a typical dinner-romantic-walk kind of proposal, but I'm sure that Alan delivered it with style. Needless to say, Terry accepted. Woo hoo! So we'll have another Squyres wedding in the next 1-2 years. Alan's a good guy; I think he'll make a great addition to the family. Now we get to meet his family (who all live in Indiana, not too far from ND, I might add!).

Rock on!

A Motley Bag o' Notes today.

• By my logs, K-Mart calls are definitely on the rise (my home phone number is the same as K-Mart's, but with 2 digits swapped). Must be because of the holiday season. The callers are also getting more and more polite -- most people are saying "sorry". Interestingly enough, there's four standard responses from people during a given K-Mart call:
  • Ok
  • Thanks
  • Sorry
  • [click] (i.e., hangup)

  I've been keeping logs on this in a flat text file. Someday I gotta hack up a PHP script and a MySQL database to do this online so that everyone can see the numbers...

• Bob from Veridan (is that what they're called these days) e-mailed me with a better solution to the PSR problem with PBS. Apparently, there's an internal PBS MOM call named fork_me() that does all the Right Things to fork a child process. This is orders of magnitude better than the popen() that we use now. Oops. I'll have to go back and fix that one up...

• Progress on poggenc is coming along swimmingly. I now pass the input .wav data all the way through the five distinct states in the state machine (input, input queue, encode, output queue, output) -- there's three separate progress bars to watch (eye candy!).

  The progress bars, themselves, turned out to be an interesting sub problem -- you want to update them all the time, but only want to actually display the value periodically. And you only need to display it if it's different than last time. But then you run into a problem when you have more than one file. So here's a sample progress line for one file:

   foo.wav            |********75%**     ||********73%**    ||********70%*      | 

  How exactly do you show the progress of multiple files (it's quite possible that multiple files are being processed simultaneously) without having to link in a curses library? Showing one line is simple -- you just output a \r instead of a \n (I think that even works in 'doze as well). But with multiple lines, without the ability to make the cursor go "up" a line, you can't do it.

  So I punted for now and just have it redisplay the whole thing again if there's more than one file being processed simultaneously. This isn't the main focus of the work, after all. :-)

  Things still to do for poggenc (in no particular order):

  1. add overlap of inputs. This is mainly a function of the input queue; need to add some extra logic to save a few readsets of the input from the tail end of every dequeue of the input and prepend them to the next dequeue.

  2. add vorbis/ogg processing.

  3. MPI stuff (only does threading for now).

  4. juice up the eye candy. If it's not worth watching, it's not worth running.
     

• xmms continues to crack me up. I think I've mentioned this before here in the journal, but there's a thread leak in it such that every song it plays launches a new thread. This thread never dies. And since in Linux threads are implemented as processes, you can see how many threads are running.

  As of right now (9:52am), I have 331 xmms processes on my Dell desktop.
  

Back to poggenc!

I have declared today "Annoying Female Vocalist Day", or AFVD, for short.

It's all queued up in xmms -- I'm set for hours of uninterrupted AFV's.

Sidenote: That is a definite benefit of digital music; you can just queue up hours and hours of music and then not have to futz with it. It's pretty much the same reason that they came out with 3- and 5-CD players. But with digital music, you can queue up [more or less] an infinite amount of music -- you're not just restricted to 3 of 5 CD's worth. There's been many a workday where I've queued up move than 18 hours of music when I start working, and then don't bother with my music for the rest of the day.

Sidenote: I love my telephone headset. It has a noise-canceling mike; I can have my music on fairly loud in the background and the person I'm talking to on the phone can't hear it at all. It's also stereo -- it has 2 earphones, contrary to most headsets. I find this to be extremely useful. It also allows sounds to be piped in from my computer -- so I can play MP3s directly over the telephone. While not an amazingly useful feature, it has been practical once or twice.

Back to AFVD.

We're starting out with Alanais Morisettet; I've got about 1.5-2 hours of her queued up; Supposed Former Infatuation Junkie and Jagged Little Pill.

Then on to Bjork -- both Telegram and Debut. That voice; my God
-- how did that happen? Hearing these albums always makes me feel like I want to donate large sums to children's charities.

We pour salt into the wound by following up with Ace of Base. True, AoB isn't completely female, but they satisfy both requirements of a) having female vocalists, and b) being annoying. We've got The Bridge and The Sign from AoB.

Then Erasure. Ok, not female at all. But he sounds female, and is definitely annoying. I Say, I Say, I Say.

The pain stays alive with Jewel -- Pieces of You and Spirit. Annoying to the max.

The agony continues with Loreena McKennit, The Book of Spirits. There's that one cool song on there that has a really bizarre-o video with some midgets running around, but other than that, it's annoying.

Nina Hagen comes next. Just ask anyone in the LSC (particularly Dog) -- Revolution Ballroom is one of the worst albums of all time. This is why you're probably never heard of her.

P.J. Harvey's Rid of Me -- so aptly titled -- prolongs the horror. I think there's one song on the album with a few quiet parts so that they can pass it off as multiple cuts to the radio stations (after all, if you're not Floyd, they're not going to play 20-30 minute songs).

Suzanne Vega's 99.9F rounds out the mix. Just like P.J., I think there's one song on this album. Consider: monotone singing with a guitar. Need I say more?

So it's going to be a Very Long day. I'll have to rely on my coding skills and embroil myself deep in hackery to keep from having my spirit crushed. It will be a true testament to my abilities if I can come out of this day and still be sane.

See you on the other side...

Just for posterity's sake, I'll make a brief journal entry.

Trip to FL to Tracy's parents for about a week was good. It was warm. We did nothing, and lots of it. I did take my laptop and spend the better portion of a day in the warm sunshine working on parallel oggenc, though.

Came back up here to a few inches of snow -- it's nothing compared to what other parts of the country are currently experiencing, but it's very unusual for Louisville to have this much snow for so long!

The last two days have been almost nothing but parallel oggenc coding/debugging. I've been pretty active on the vorbis-dev list since I got back; much discussion has occurred. Monty tells me that he's got a branch where the thread-safe encoding stuff is partially done (whoo hoo!!!); can't wait for that to become mainstream.

I've gotten pretty far with poggenc, and it appears to almost be working. Things yet to do:

• Output ogg data appears to be getting hosed somewhere in the output queue. It varies from completely bad output to occasional "blips" in the output audio, but I think it's symptoms of the same problem.

• MPI stuff hasn't been written yet; it's all threaded right now.

• My stats displays will have to be re-thunk a bit. Right now, there's 3 separate displays (done that way on purpose; one for input, encoding, and output). But it might make it easier to have a single object for all three (rather than three objects) because I think I want to have an "oggenc compatibility mode" stats display where there's only one progress bar so that programs like grip can run poggenc without having to analyze a different progress bar.

• All the steam shutdown bookkeeping isn't written yet; I don't think there's much, but I'm concentrating on getting it working before I make all the memory cleanup at the end work properly.
  

The first item is the focus of my current work; it's been a bear to find some far -- I can't manage to track it down somehow. We'll see...

Heading out to a New Year's party later tonight which should be fun. I'll probably be up at ND later this week (not for sure yet, but likely).

Special shout out to my homie Suzanne -- happy birthday!